CAIDE Score,Alzheimer’s Disease Pathology,and Cognition in Cognitively Normal Adults: The CABLE Study

Abstract

Background:

Cardiovascular Risk Factors, Ageing and Dementia (CAIDE) risk score serves as a credible predictor of an individual’s risk of dementia. However, studies on the link of the CAIDE score to Alzheimer’s disease (AD) pathology are scarce.

Objective:

To explore the links of CAIDE score to cerebrospinal fluid (CSF) biomarkers of AD as well as to cognitive performance.

Methods:

In the Chinese Alzheimer’s Biomarker and LifestylE (CABLE) study, we recruited 600 cognitively normal participants. Correlations between the CAIDE score and CSF biomarkers of AD as well as cognitive performance were probed through multiple linear regression models. Whether the correlation between CAIDE score and cognitive performance was mediated by AD pathology was researched by means of mediation analyses.

Results:

Linear regression analyses illustrated that CAIDE score was positively associated with tau-related biomarkers, including pTau (p < 0.001), tTau (p < 0.001), as well as tTau/Aβ42 (p = 0.008), while it was in negative association with cognitive scores, consisting of MMSE score (p < 0.001) as well as MoCA score (p < 0.001). The correlation from CAIDE score to cognitive scores was in part mediated by tau pathology, with a mediation rate varying from 3.2% to 13.2%.

Conclusions:

A higher CAIDE score, as demonstrated in our study, was linked to more severe tau pathology and poorer cognitive performance, and tau pathology mediated the link of CAIDE score to cognitive performance. Increased dementia risk will lead to cognitive decline through aggravating neurodegeneration.

Keywords

Alzheimer’s disease CAIDE dementia risk score cerebrospinal fluid biomarkers cognition neurodegeneration

INTRODUCTION

Alzheimer’s disease (AD), commonly known as dementia, accounts for the largest proportion of dementia [1]. This disease is accompanied by progressive cognitive decline and leads to loss of self-care eventually [2]. AD primarily affects the elderly population, and the incidence of AD increases with population ageing, imposing on individuals as well as on society substantial burdens [3]. However, an effective treatment for AD remains elusive [4]. Therefore, the prediction and prevention of this disease are of great significance.

Previous studies have confirmed the strong association between cardiovascular diseases and AD [5 –7]. Based on this fact, models consisting of cardiovascular risk factors to forecast dementia risk are constantly being devised [8 –10]. Of these, the Cardiovascular Risk Factors, Ageing and Dementia (CAIDE) risk score serves as a credible predictor of an individual’s risk of dementia [11], as confirmed by preceding longitudinal studies in different races [12 –14]. The CAIDE score consists of a series of cardiovascular risk factors and dementia risk factors, including age, sex, education, systolic blood pressure (SBP), total cholesterol (TC), body mass index (BMI), and physical activity (PA). The information needed to calculate the CAIDE score is more readily available than in other risk models [8 –11], and it was commonly used to assess dementia risk in previous studies [15 –17]. Prior studies have demonstrated an increased risk of dementia, as determined by the CAIDE score, was closely tied to worse objective cognitive functioning [18, 19]. However, studies on the link of CAIDE score to AD pathology are scarce.

Considering that AD pathology may exist before the onset of cognitive impairment [20, 21] and that AD pathological development correlates well with cognitive decline [4], examining the link of CAIDE score to AD pathology will contribute to our understanding of the potential mechanisms by which elevated dementia risk leads to cognitive decline. Therefore, in this research, we intended to (1) delve into the connection between CAIDE score and cerebrospinal fluid (CSF) biomarkers of AD, (2) validate the link of CAIDE score to cognitive performance, and (3) investigate whether AD pathology mediates the correlation between CAIDE score and cognitive performance.

METHODS

Participants

We recruited participants from the Chinese Alzheimer’s Biomarker and LifestylE (CABLE) study [22], which began in 2017 with the primary aim of identifying genetic and lifestyle risk factors linked to AD biomarkers. Participants were all Han Chinese hospitalized in Qingdao Municipal Hospital of China, between 40 and 90 years of age. We excluded those with (1) primary neurological diseases such as epilepsy and central nervous system infections; (2) systemic diseases that may affect AD biomarkers such as malignant tumors; (3) hereditary diseases; or (4) a history of previous lumbar spine surgery. All the participants underwent blood and CSF sampling, and assessments of cognitive performance. Demographic information was collected through the hospital information system and questionnaires. An application was submitted to the Ethics Committee and received its approval. All the participants signed informed consent.

CAIDE risk score

There are two versions of the CAIDE risk score. Version 1 includes seven components: age, sex, education, SBP, TC, BMI, and PA [11], while version 2 additionally includes APOE ɛ4 status. The specific scoring methods are shown in SupplementaryTable 1. Since the inclusion of APOE ɛ4 status does not improve the accuracy of the CAIDE score in predicting dementia [11] and information about APOE ɛ4 status is not universally available, we chose version 1 of the CAIDE score for our analyses to improve the generalizability and reproducibility of the results, while version 2 will serve as a sensitivity analysis. A score of 9 was considered to be the best cutoff for version 1 of the CAIDE score with the best sensitivity and specificity when predicting dementia [11], which was also widely validated by previous studies in different races [15 , 24]. Therefore, we selected this cutoff of 9 to divide all the participants into a high-risk group (CAIDE score≥9) and a low-risk group (CAIDE score < 9) to perform comparisons of demographic information, CSF biomarkers of AD, and cognitive scores.

CSF biomarkers of AD

CSF was collected by lumbar puncture and stored in a freezer at –80°C. Aβ₄₂, pTau, and tTau were assayed using ELISA kits (Innotest; Fujirebio, Ghent, Belgium). Previous studies have demonstrated that AD pathological transformation occurs in nearly one-third of cognitively intact older individuals [20, 21]. Therefore, we chose the one-third quartile of CSF AD biomarker levels of the total population as cutoffs to distinguish normal from abnormal CSF AD biomarker levels: <177.80 pg/ml (less than one-third of the overall level) for Aβ₄₂, >35.84 pg/ml (greater than one-third of the overall level) for pTau, and >144.97 pg/ml (greater than one-third of the overall level) for tTau. In addition, given that preclinical stage was a critical period of dementia prevention [25], we further divided the participants into three subgroups: the healthy control (HC) group (normal Aβ₄₂, pTau, and tTau; n = 84), the preclinical AD group (abnormal Aβ₄₂; n = 200), and the suspected non-AD pathology group (SNAP) (normal Aβ₄₂ but abnormal pTau or tTau; n = 316) on the basis of the National Institute on Aging-Alzheimer’s Association (NIA-AA) criteria [26 –29].

Cognitive performance

The cognitive performance was evaluated via the Mini-Mental State Examination (MMSE) [30] as well as the Montreal Cognitive Assessment (MoCA) [31]. The thresholds for the MMSE score are defined as follows:≤17 for 0 years of schooling, ≤20 for ≤6 years of schooling, and ≤24 for >6 years of schooling. The MoCA score thresholds are <19 for <7 years of schooling, <22 for years of schooling ≥7 but <24, and <24 for >12 years of schooling.

APOE ɛ4 status

Fasting blood samples were centrifuged and then stored in a freezer at –20°C. The DNA in the samples was retrieved utilizing the QIAamp^® DNA Blood Mini-Kit (250). The status of rs7412 and rs429358, two variants defining APOE status, were determined through restriction fragment length polymorphism (RFLP) techniques. Participants were categorized into APOE ɛ4 carriers and APOE ɛ4 non-carriers according to their APOE ɛ4 carriage.

Statistical analyses

A total of 1,730 participants who were cognitively normal according to MMSE score were recruited from the CABLE study, of whom those who did not have the information needed to calculate the CAIDE score, information on CSF biomarkers, as well as information of other lifestyle and co-morbidities, were excluded, which excluded 1121 participants. Participants with CSF biomarker levels outside four times the standard deviation of the overall level were also excluded, which further excluded 9 participants. Finally, 600 participants were included in this cross-sectional study. Detailed participants screening process were showed in Supplementary Figure 1.

The pTau/Aβ₄₂ and tTau/Aβ₄₂ ratios were better predictors of cognitive performance than these biomarkers alone [32]. Therefore, we included them as dependent variables in our analyses. Considering the skewed distribution of the data for the biomarkers and cognitive scores, we performed a Box-Cox transformation [33] on them by means of the bcPower function in the “car” package in the 4.2.1 version of the R software to bring them closer to a normal distribution.

The demographic characteristics of the group at low risk, including the information needed to calculate the CAIDE score, other lifestyles and co-morbidities, and proportion of rural population, were compared to those of the group at high risk. For binary qualitative variables’ comparisons, we utilized the Chi-square test, whereas for continuous variables’ comparisons, we utilized the Mann-Whitney U test or t-test. Linear regression analyses were utilized to explore the link of CAIDE score (continuous) to CSF biomarkers of AD. In addition, the connection from CAIDE scores to CSF AD biomarkers abnormality was investigated utilizing logistic regression analyses. The link of the CAIDE score to cognitive scores was validated via linear regression models. Subgroup analyses by AD pathology (HC, preclinical AD or SNAP), age (mid-life (less than 65 years) and late-life (greater than or equal to 65 years)), and APOE status (with or without APOE ɛ4) were performed to validate the link of CAIDE score to biomarkers.

After verifying the significant correlation between (1) CAIDE score and cognitive scores, (2) CAIDE score and biomarkers, and (3) biomarkers and cognitive scores. To figure out whether the link of CAIDE score to cognitive performance was mediated by AD pathology, according to the approach of Preacher and Hayes [34], mediation analyses were conducted. The Sobel test [35] and the 95% confidence intervals (CI) it produces in its application were used to evaluate the significance of the mediating effect. The mediating effect was significant if the 95% CI did not include zero.

Sensitivity analyses were conducted using version 2 of the CAIDE score. Unless otherwise stated, multiple corrections were made following the Bonferroni method. Data analyses and graphs were performed using version 4.2.1 of the R software and version 8.0 of the GraphPad Prism software.

RESULTS

Population characteristics

Table 1 demonstrates the results of group comparisons. The participants’ age was 60.21 years on average (SD = 10.84), of which 40.2% were women. The participants’ mean CAIDE score was 6.83 (SD = 2.44). The high-risk group included 141 participants, accounting for 23.5% of the total population. They were older, more likely to be female, less educated, had more cardiovascular risk factors, and more from rural area, compared to the low-risk group (p < 0.05).

Table 1

Population characteristics

Characteristics	CAIDE risk score categories		Total	p
	Low-risk	High-risk
N	459	141	600
Age (y, mean±SD)	58.61±11.16	65.40±7.73	60.21±10.84	<0.001
Sex (female, yes, %)	172 (37.5)	69 (48.9)	241 (40.2)	0.020
Education (y, mean±SD)	11.33±3.77	7.24±3.36	10.37±4.07	<0.001
SBP>140 mmHg (yes, %)	106 (23.1)	108 (76.6)	214 (35.7)	<0.001
TC (mmol/l, mean±SD)	4.82±1.02	5.34±1.34	4.94±1.13	<0.001
BMI (kg/m², mean±SD)	25.01±3.35	27.11±4.48	25.50±3.75	<0.001
PA (yes, %)	214 (46.6)	77 (54.6)	291 (48.5)	0.118
APOE ɛ4 carriers (yes, %)	71 (15.5)	20 (14.2)	91 (15.2)	0.812
Smoking (yes, %)	146 (31.8)	37 (26.2)	183 (30.5)	0.25
Alcohol (yes, %)	87 (19.0)	20 (14.2)	107 (17.8)	0.243
DM (yes, %)	56 (12.2)	31 (22.0)	87 (14.5)	0.006
CHD (yes, %)	53 (11.5)	18 (12.8)	71 (11.8)	0.808
Area (countryside, yes, %)	75 (16.3)	49 (34.8)	124 (20.7)	<0.001
CSF biomarkers of AD
Aβ₄₂ (pg/ml, mean±SD)	270.65±157.27	270.63±159.94	270.65±157.76	0.931
pTau (pg/ml, mean±SD)	41.12±13.20	46.77±14.55	42.45±13.73	<0.001
tTau (pg/ml, mean±SD)	185.82±81.08	218.35±90.04	193.46±84.34	<0.001
pTau/Aβ₄₂ (pg/ml, mean±SD)	0.22±0.17	0.24±0.17	0.22±0.17	0.030
tTau/Aβ₄₂ (pg/ml, mean±SD)	0.96±0.79	1.11±0.90	0.99±0.82	0.010
Cognitive scores
MMSE score (mean±SD)	28.64±1.56	27.71±2.18	28.42±1.77	<0.001
MoCA score (mean±SD)	25.86±2.49	23.81±2.62	25.40±2.65	<0.001

Comparisons between groups were using chi-square test for binary qualitative data, t-test for normalized continuous variables, and Mann–Whitney U test for non-normalized variables. p < 0.05 were considered significant. Significant differences were in bold. SD, standard deviation; SBP, systolic blood pressure; TC, total cholesterol; BMI, body mass index; PA, physical activity; DM, diabetes mellitus; CHD, coronary heart disease; CSF, cerebrospinal fluid; AD, Alzheimer’s disease; MMSE, Mini-Mental State Examination; MoCA, Montreal Cognitive Assessment.

CAIDE score and CSF biomarkers of AD

The high-risk group had more severe tau pathology, consisting of pTau (p < 0.001), tTau (p < 0.001), pTau/Aβ₄₂ (p = 0.030), as well as tTau/Aβ₄₂ (p = 0.010), compared to low-risk group (Fig. 1). However, there was no difference in Aβ₄₂ levels of the group at high risk compared to the group at low risk. Linear regression analyses showed that CAIDE score was positively correlated with tau-related biomarkers, including pTau (p < 0.001), tTau (p < 0.001), as well as tTau/Aβ₄₂ (p = 0.003) (Supplementary Table 2), the result did not change after adjusting for APOE ɛ4 status (with or without APOE ɛ4), smoking (yes or no), drinking (yes or no), diabetes mellitus (with or without), as well as coronary heart disease (with or without) (Supplementary Table 2 and Fig. 2). No association was observed between the CAIDE score and Aβ₄₂. In logistic regression analyses, a higher CAIDE score was associated with pTau abnormality (p < 0.001) and tTau abnormality (p < 0.001) but not with Aβ abnormality (p = 0.093) after adjusting for APOE ɛ4 status, smoking, alcohol consumption, diabetes mellitus, as well as coronary heart disease (Supplementary ).

Fig. 1

CSF biomarkers of AD in different risk groups. Comparisons were made using t-test. p < 0.05 was considered significant. Significant results were in bold. CSF, cerebrospinal fluid.

Fig. 2

Correlation between CAIDE score and CSF biomarkers of AD. Multiple linear regression models were used to explore the correlation between CAIDE score and CSF biomarkers of AD, adjusting for APOE status (with or without APOE ɛ4), smoking (yes or no), drinking (yes or no), diabetes mellitus (yes or no), and coronary heart disease (yes or no). p < 0.05/5 was considered significant. Significant results were in bold. CSF, cerebrospinal fluid.

CAIDE score and cognitive scores

Cognitive scores of the group at high risk were found to be lower than those of the group at low risk (p < 0.001) (Table 1). Linear regression analyses revealed that higher CAIDE risk score correlates well with lower cognitive scores consisting of MMSE score (p < 0.001) as well as MoCA score (p < 0.001), the result did not change after adjusting for APOE ɛ4 status, smoking, drinking, diabetes mellitus, as well as coronary heart disease (p < 0.001 for MMSE score as well as MoCA score) (Table 2).

Table 2

Correlation between CAIDE score and cognitive scores

	MMSE score		MoCA score
	β	p	β	p
Model 1	–0.115	<0.001	–0.183	<0.001
Model 2	–0.106	<0.001	–0.185	<0.001

Linear regression models were conducted to explore the correlation between CAIDE score and cognitive scores. Model 1: without covariate. Model 2: adjusted for APOE ɛ4 status (with or without APOE ɛ4), smoking (yes or no), drinking (yes or no), diabetes mellitus (with or without), as well as coronary heart disease (with or without). p < 0.05/2 were considered significant. Significant results were in bold. MMSE, Mini-Mental State Examination; MoCA, Montreal Cognitive Assessment.

Stratified analyses by AD pathology, age, and APOE ɛ4 status

In subgroups by AD pathology, we found that only in the preclinical AD and SNAP groups were correlations between CAIDE score and tau pathology observed. This was not the case in the HC group. In the preclinical AD group, a higher CAIDE score was correlated with higher pTau and tTau, though this result did not survive multiple corrections (p = 0.015 for pTau and p = 0.011 for tTau) (Fig. 3). Also, in the SNAP group, a higher CAIDE score was linked to all tau-related biomarkers, including pTau (p < 0.001), tTau (p < 0.001), pTau/Aβ₄₂ (p = 0.001), as well as tTau/Aβ₄₂ (p < 0.001) (Fig. 3). However, in the HC group, we did not find any correlation between the CAIDE score and biomarkers. Additionally, only in participants at mid-life can we observe a link between CAIDE score and tau-related biomarkers, including pTau (p < 0.001) as well as tTau (p < 0.001), not in participants at late-life (Fig. 3). The connection between CAIDE score and cerebrospinal fluid (CSF) biomarkers of AD did not differ significantly in subgroups of different APOE ɛ4 status.

Fig. 3

Subgroup analyses of the correlation between CAIDE score and CSF biomarkers of AD. Multiple linear regression models were used to explore the correlation between CAIDE score and biomarkers, adjusting for APOE status (with or without APOE ɛ4), smoking (yes or no), drinking (yes or no), diabetes mellitus (yes or no) and coronary heart disease (yes or no). p < 0.05/5 was considered significant. HC, healthy control; AD, Alzheimer’s disease; SNAP, suspect non-AD pathology. Red color represents a positive correlation between CAIDE score and CSF biomarkers of AD and blue color represents a negative correlation between CAIDE score and CSF biomarkers of AD. Shades of color represent the magnitude of the absolute value of the β coefficient in the multiple linear regression analysis. ^*p < 0.05/5; ^**p < 0.01/5; ^***p < 0.001/5; ^****p < 0.0001/5.

Mediation analyses

The correlation between CAIDE score and cognitive scores was in part mediated by tau pathology. The correlation between CAIDE score and MMSE score was mediated by pTau (mediation ratio = 12.3%; 95% CI: 8.1% –17.0%; p = 0.001), tTau (mediation ratio = 13.2%; 95% CI: 8.1% –17.5%; p = 0.005) as well as tTau/Aβ₄₂ (mediation ratio = 8.6%; 95% CI: 4.1% –13.2%; p = 0.024) (Fig. 4A–C). Consistently, tTau and tTau/Aβ₄₂ mediated the correlation between CAIDE score and MoCA score, with a mediation ratio of 5.9% for tTau (95% CI: 1.9% –9.7%; p = 0.007) and 3.2% for tTau/Aβ₄₂ (95% CI: 1.3% –6.1%; p = 0.046) (Fig. 4D, E).

Fig. 4

Mediation analysis of CSF biomarkers between CAIDE score and cognitive scores. Mediation analyses were performed using the non-parametric bootstrapping approach of Preacher and Hayes to examine the mediation effect of tau pathology on the correlation between CAIDE score and cognitive scores. pTau, tTau, and tTau/Aβ₄₂ mediated the relationship between CAIDE score and MMSE score (A-C), and tTau and tTau/Aβ₄₂ mediated the relationship between CAIDE score and MoCA score (D and E). Each model path was adjusted for APOE status (with or without APOE ɛ4), smoking (yes or no), drinking (yes or no), diabetes mellitus (yes or no), and coronary heart disease (yes or no). MMSE, Mini-Mental State Examination; MoCA, Montreal Cognitive Assessment.

Extra validation of version 2 of the CAIDE score

Consistent with the results obtained from analyses using version 1 of CAIDE score, linear regression analyses shown that, after adjusting for smoking, drinking, diabetes mellitus, and coronary heart disease, version 2 of CAIDE score was positively associated with tau-related biomarkers, including pTau (p < 0.001), tTau (p < 0.001), as well as tTau/Aβ₄₂ (p = 0.003) (Supplementary Table 4), while it was in negative association with cognitive scores, including MMSE score (β= –0.093, p < 0.001) as well as MoCA score (β= –0.155, p < 0.001). Correlation between CAIDE score and MMSE score was mediated by pTau (mediation ratio = 9.7%; 95% CI: 5.9% –14.4%; p = 0.003), tTau (mediation ratio = 11.8%; 95% CI: 6.1% –14.2%; p = 0.006) and tTau/Aβ₄₂ (mediation ratio = 8.6%; 95% CI: 4.5% –13.6%; p = 0.008) (Supplementary Figure 2A–C), and tTau and tTau/Aβ₄₂ mediated the correlation between CAIDE score and MoCA score, with a mediation ratio of 5.2% for tTau (95% CI: 1.5% –9.0%; p = 0.046) and 3.2% for tTau/Aβ₄₂ (95% CI: 1.5% –6.1%; p = 0.012) (Supplementary D, E).

DISCUSSION

Our study showed that a higher CAIDE score was strongly linked to more severe tau pathology and poorer cognitive performance. In addition, the link of CAIDE score to cognitive performance was in part mediated by tau pathology. These results suggest a role for tau pathology in the process of cognitive decline due to elevated dementia risk.

Regarding amyloid, we did not find a link between CAIDE score and amyloid in various populations, following the results of several previous studies [36 –38]. Nevertheless, two studies reported different results from ours. The first study included 724 participants who had cognitive impairment. Thresholds for the CAIDE score were set at values that would roughly average the participants into three groups: low-risk, high-risk, and middle-risk groups. Then a positive association between risk level and Aβ₄₂ (measured using commercially available sandwich enzyme-linked immunoabsorbent assays) was indicated [39]. Comparing this study with ours, we proposed several possible causes for this inconsistency. Firstly, all participants in this study were reported to have cognitive problems and may come from a more heterogeneous group. Secondly, the grouping methods and biomarker assays of this study differed from those of ours. In the second study, a positive correlation between CAIDE score and Aβ-PET positivity was reported [18]. After comparing it to our study, we suggested that different sample sizes and amyloidosis assessment modalities might explain for the inconsistent results. This study recruited only 159 participants, including participants with dementia, which will lead to greater error. Second, they used PET to measure brain amyloid load, which in a cognitively normal population may give results different from those obtained by CSF biomarker assays [40].

About tau-related biomarkers, only one study has systematically investigated their relationships with CAIDE score, which happens to be the same study that found the correlation between CAIDE score and Aβ₄₂. This study reported that the CAIDE score was associated with all tau-related biomarkers except pTau [39]. Whereas in our study, the CAIDE score was also associated with pTau and pTau abnormality. The possible reasons for this inconsistency may be related to the different CSF biomarker assays and grouping methods. CAIDE score was positively associated with pTau and tTau in the preclinical AD group, but not in the HC group, suggesting that increased dementia risk will lead to neurodegeneration in the initial phase of AD. Additionally, higher CAIDE score was associated with tau-related biomarkers, including pTau, tTau, pTau/Aβ₄₂, and tTau/Aβ₄₂ in the SNAP group, implying that increased dementia risk was associated with neuronal damage in other neurodegenerative diseases apart from AD, a relationship that needs to be studied further. The association between CAIDE score and tau pathology was more significant in SNAP group than in preclinical AD group, this might be attributed to the larger sample size of SNAP group. Stratified analyses by age showed that CAIDE score was positively associated with pTau and tTau in participants at mid-life, but not in those at late-life, emphasizing the importance of dementia risk reduction in middle-aged populations to delay neurodegeneration.

Most of the components in CAIDE score are closely related to tau pathology, especially age, SBP, and APOE ɛ4. Intracellular and extracellular age-related pathophysiological changes, such as DNA methylation and nutrient sensing deregulation, could exacerbate tau protein phosphorylation [41]. Abnormal blood pressure might cause nerve damage and decreased clearance of AD pathological proteins through reducing cerebral blood flow [42] and damage blood-brain barrier [43, 44]. In addition, APOE ɛ4 expressed by neurons could aggravate tau hyperphosphorylation [45].

Regarding cognitive performance, previous studies have found that the CAIDE score adversely affects cognitive performance [18, 19]. In our research, based on a larger population, we found similar results. Tau protein correlates well with cognitive decline [46], with pTau initiating the process of neurodegeneration [4]. In our study, we found that tau pathology mediated the link of CAIDE score to cognitive scores, implying that increased dementia risk might contribute to cognitive decline through aggravating neurodegeneration.

The relationship between CAIDE score and cognition could be mainly attributed to several factors in CAIDE score, including age, education, SBP, and PA. Previous study have confirmed that age and education were closed related to cognition [47 –49], in which education could improve cognitive and brain reserve, improving the brain tolerance to pathology and allowing the use of pre-existing compensatory approaches to cope with brain damage [50]. Hypertension could intensify pathophysiological processes related to brain damage, including endothelial activation, inflammation, and ischemic damage[51], while regular exercise increases the supplement of blood and oxygen in brain and reduces the risk of chronic diseases related to cognitive decline [52].

In this study, we conducted analyses in individuals who were cognitively normal, further validating the predictive ability of the CAIDE score. However, there were some limitations. Firstly, we only performed the analyses in the Chinese Han population, we didn’t compare the differences between races, in which these associations may change. Secondly, the cognitive screening tools used in our study, including the MMSE score and MoCA score, had lower sensitivity and specificity when distinguishing between cognitive normalcy and cognitive impairment compared to comprehensive neuropsychologic tests [53], which would lead to more errors in the results. In addition, all the analyses were cross-sectional and cannot explain a causal relationship. Future studies should validate the link of CAIDE score to AD pathology among different races based on longitudinal data.

Conclusions

AUTHOR CONTRIBUTIONS

Ze-Xin Guo (Data curation; Formal analysis; Visualization; Writing – original draft); Fang Liu (Formal analysis); Fang-Yuan Wang (Data curation; Formal analysis; Visualization; Writing – review & editing); Ya-Nan Ou (Data curation; Formal analysis; Visualization; Writing – review & editing); Liang-Yu Huang (Data curation; Formal analysis; Visualization; Writing – review & editing); Hao Hu (Data curation; Writing – review & editing); Zhi-Bo Wang (Data curation; Writing – review & editing); Yan Fu (Data curation; Writing – review & editing); Pei-Yang Gao (Data curation; Writing – review & editing); Lan Tan (Conceptualization; Data curation; Project administration; Resources; Writing – review & editing); Jin-Tai Yu (Conceptualization; Funding acquisition; Project administration; Resources; Writing – review & editing).

Footnotes

ACKNOWLEDGMENTS

Thanks to the planners and executives who worked on the CABLE study.

FUNDING

This study was supported by grants from the National Natural Science Foundation of China (82071201 and 81971032), Science and Technology Innovation 2030 Major Projects (2022ZD0211600), Shanghai Municipal Science and Technology Major Project (No.2018SHZDZX01), Research Start-up Fund of Huashan Hospital (2022QD002), Excellence 2025 Talent Cultivation Program at Fudan University (3030277001), and ZHANGJIANG LAB, Tianqiao and Chrissy Chen Institute, and the State Key Laboratory of Neurobiology and Frontiers Center for Brain Science of Ministry of Education, Fudan University.

CONFLICT OF INTEREST

Jin-Tai Yu and Lan Tan are Editorial Board Members of this journal but were not involved in the peer-review process of this article nor had access to any information regarding its peer-review.

DATA AVAILABILITY

Data used in the current study can be obtained from the corresponding author if a reasonable request is made.

The supplementary material is available in the electronic version of this article:.

References

Atri

(2019) The Alzheimer’s disease clinical spectrum: Diagnosis and management. Med Clin North Am 103, 263–293.

Albert

(2011) Changes in cognition. Neurobiol Aging 32,(Suppl 1) S58–63.

(2023) 2023 Alzheimer’s disease facts and figures. Alzheimers Dement 19, 1598–1695.

Long

, Holtzman

(2019) Alzheimer disease: An update on pathobiology and treatment strategies. Cell 179, 312–339.

Eisenmenger

, Peret

, Famakin

, Spahic

, Roberts

, Bockholt

, Johnson

, Paulsen

(2023) Vascular contributions to Alzheimer’s disease. Transl Res 254, 41–53.

Lespinasse

, Chêne

, Mangin

, Dubois

, Blanc

, Paquet

, Hanon

, Planche

, Gabelle

, Ceccaldi

, Annweiler

, Krolak-Salmon

, Godefroy

, Wallon

, Sauvée

, Bergeret

, Chupin

, Proust-Lima

, Dufouil

(2022) Associations among hypertension, dementia biomarkers, and cognition: The MEMENTO cohort. Alzheimers Dement 19, 2332–2342.

Janka

, Juhász

, Rimanóczy

, Boda

, Márki-Zay

, Kálmán

(2002) Codon 311 (Cys–>Ser) polymorphism of paraoxonase-2 gene is associated with apolipoprotein E4 allele in both Alzheimer’s and vascular dementias. Mol Psychiatry 7, 110–112.

Licher

, Leening

MJG

, Yilmaz

, Wolters

, Heeringa

, Bindels

PJE

, Vernooij

, Stephan

BCM

, Steyerberg

, Ikram

(2019) Development and validation of a dementia risk prediction model in the general population: An analysis of three longitudinal studies. Am J Psychiatry 176, 543–551.

Anstey

, Cherbuin

, Herath

, Qiu

, Kuller

, Lopez

, Wilson

, Fratiglioni

(2014) A self-report risk index to predict occurrence of dementia in three independent cohorts of older adults: The ANU-ADRI. PLoS One 9, e86141.

10.

Schiepers

OJG

, Köhler

, Deckers

, Irving

, O’Donnell

, van den Akker

, Verhey

FRJ

, Vos

SJB

, de Vugt

, van Boxtel

MPJ

(2018) Lifestyle for Brain Health (LIBRA): A new model for dementia prevention. Int J Geriatr Psychiatry 33, 167–175.

11.

Kivipelto

, Ngandu

, Laatikainen

, Winblad

, Soininen

, Tuomilehto

(2006) Risk score for the prediction of dementia risk in 20 years among middle aged people: A longitudinal, population-based study. Lancet Neurol 5, 735–741.

12.

Exalto

, Quesenberry

, Barnes

, Kivipelto

, Biessels

, Whitmer

(2014) Midlife risk score for the prediction of dementia four decades later. Alzheimers Dement 10, 562–570.

13.

Virta

, Heikkilä

, Perola

, Koskenvuo

, Räihä

, Rinne

, Kaprio

(2013) Midlife cardiovascular risk factors and late cognitive impairment. Eur J Epidemiol 28, 405–416.

14.

Hou

, Feng

, Zhang

, Cao

, Tan

, Yu

(2019) Models for predicting risk of dementia: A systematic review. J Neurol Neurosurg Psychiatry 90, 373–379.

15.

Wittmann

, Zülke

, Pabst

, Luppa

, Thyrian

, Kästner

, Hoffmann

, Kaduszkiewicz

, Döhring

, Escales

, Gensichen

, Zöllinger

, Kosilek

, Wiese

, Oey

, König

, Brettschneider

, Frese

, Riedel-Heller

(2023) COVID-19 government measures and their impact on mental health: A cross-sectional study of older primary care patients in Germany. Front Public Health 11, 1141433.

16.

Hua

, Xiong

, Li

, Zhu

, Ge

, Ma

, Fu

, Li

, Wang

, Dong

, Zhao

, Ma

, Chen

, Gao

, He

, Wang

, Wei

, Wang

, Gao

, Chen

, Zeng

, Xie

(2022) Development and validation of a deep learning algorithm based on fundus photographs for estimating the CAIDE dementia risk score. Age Ageing 51, afac282.

17.

Shah

, Ysea-Hill

, Torres-Morales

, Gomez

, Castellanos

, Ruiz

(2022) Identifying dementia risk in older veterans using a mailing survey. J Prev Alzheimers Dis 9, 371–375.

18.

Lockhart

, Schaich

, Craft

, Sachs

, Rapp

, Jung

, Whitlow

, Solingapuram Sai

, Cleveland

, Williams

, Burke

, Bertoni

, Hayden

, Hughes

(2022) Associations among vascular risk factors, neuroimaging biomarkers, and cognition: Preliminary analyses from the Multi-Ethnic Study of Atherosclerosis (MESA). Alzheimers Dement 18, 551–560.

19.

Iso-Markku

, Kaprio

, Lindgren

, Rinne

, Vuoksimaa

(2021) Middle-age dementia risk scores and old-age cognition: A quasi-experimental population-based twin study with over 20-year follow-up. J Neurol Neurosurg Psychiatry 92, 323–330.

20.

Rowe

, Ellis

, Rimajova

, Bourgeat

, Pike

, Jones

, Fripp

, Tochon-Danguy

, Morandeau

, O’Keefe

, Price

, Raniga

, Robins

, Acosta

, Lenzo

, Szoeke

, Salvado

, Head

, Martins

, Masters

, Ames

, Villemagne

(2010) Amyloid imaging results from the Australian Imaging, Biomarkers and Lifestyle (AIBL) study of aging. Neurobiol Aging 31, 1275–1283.

21.

Ingala

, De Boer

, Masselink

, Vergari

, Lorenzini

, Blennow

, Chételat

, Di Perri

, Ewers

, van der Flier

, Fox

, Gispert

, Haller

, Molinuevo

, Muniz-Terrera

, Mutsaerts

, Ritchie

, Schmidt

, Schwarz

, Vermunt

, Waldman

, Wardlaw

, Wink

, Wolz

, Wottschel

, Scheltens

, Visser

, Barkhof

(2021) Application of the ATN classification scheme in a population without dementia: Findings from the EPAD cohort. Alzheimers Dement 17, 1189–1204.

22.

Hou

, Bi

, Tan

, Xu

, Li

, Shen

, Dou

, Tan

, Yu

(2019) Genome-wide association study identifies Alzheimer’s risk variant in MS4A6A influencing cerebrospinal fluid sTREM2 levels. Neurobiol Aging 84, 241.e213–241.e220.

23.

Zülke

, Luppa

, Röhr

, Weißenborn

, Bauer

, Samos

, Kühne

, Zöllinger

, Döhring

, Brettschneider

, Oey

, Czock

, Frese

, Gensichen

, Haefeli

, Hoffmann

, Kaduszkiewicz

, König

, Thyrian

, Wiese

, Riedel-Heller

(2021) Association of mental demands in the workplace with cognitive function in older adults at increased risk for dementia. BMC Geriatr 21, 688.

24.

Ecay-Torres

, Estanga

, Tainta

, Izagirre

, Garcia-Sebastian

, Villanua

, Clerigue

, Iriondo

, Urreta

, Arrospide

, Díaz-Mardomingo

, Kivipelto

, Martinez-Lage

(2018) Increased CAIDE dementia risk, cognition, CSF biomarkers, and vascular burden in healthy adults. Neurology 91, e217–e226.

25.

Dubois

, Hampel

, Feldman

, Scheltens

, Aisen

, Andrieu

, Bakardjian

, Benali

, Bertram

, Blennow

, Broich

, Cavedo

, Crutch

, Dartigues

, Duyckaerts

, Epelbaum

, Frisoni

, Gauthier

, Genthon

, Gouw

, Habert

, Holtzman

, Kivipelto

, Lista

, Molinuevo

, O’Bryant

, Rabinovici

, Rowe

, Salloway

, Schneider

, Sperling

, Teichmann

, Carrillo

, Cummings

, Jack

, Jr. (2016) Preclinical Alzheimer’s disease: Definition, natural history, and diagnostic criteria. Alzheimers Dement 12, 292–323.

26.

Sperling

, Aisen

, Beckett

, Bennett

, Craft

, Fagan

, Iwatsubo

, JackCR,Jr., , Kaye

, Montine

, Park

, Reiman

, Rowe

, Siemers

, Stern

, Yaffe

, Carrillo

, Thies

, Morrison-Bogorad

, Wagster

, Phelps

(2011) Toward defining the preclinical stages of Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7, 280–292.

27.

JackCR,Jr., , Knopman

, Weigand

, Wiste

, Vemuri

, Lowe

, Kantarci

, Gunter

, Senjem

, Ivnik

, Roberts

, Rocca

, Boeve

, Petersen

(2012) An operational approach to National Institute on Aging-Alzheimer’s Association criteria for preclinical Alzheimer disease. Ann Neurol 71, 765–775.

28.

Albert

, DeKosky

, Dickson

, Dubois

, Feldman

, Fox

, Gamst

, Holtzman

, Jagust

, Petersen

, Snyder

, Carrillo

, Thies

, Phelps

(2011) The diagnosis of mild cognitive impairment due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7, 270–279.

29.

McKhann

, Knopman

, Chertkow

, Hyman

, JackCR,Jr., , Kawas

, Klunk

, Koroshetz

, Manly

, Mayeux

, Mohs

, Morris

, Rossor

, Scheltens

, Carrillo

, Thies

, Weintraub

, Phelps

(2011) The diagnosis of dementia due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7, 263–269.

30.

Jia

, Wang

, Huang

, Su

, Du

, Jiang

, Wang

, Su

, Xiao

, Wang

, Zhang

(2021) A comparison of the Mini-Mental State Examination (MMSE) with the Montreal Cognitive Assessment (MoCA) for mild cognitive impairment screening in Chinese middle-aged and older population: A cross-sectional study. BMC Psychiatry 21, 485.

31.

Chen

, Xu

, Chu

, Ding

, Liang

, Nasreddine

, Dong

, Hong

, Zhao

, Guo

(2016) Validation of the Chinese Version of Montreal Cognitive Assessment Basic for Screening Mild Cognitive Impairment. J Am Geriatr Soc 64,e285–e290.

32.

Racine

, Koscik

, Nicholas

, Clark

, Okonkwo

, Oh

, Hillmer

, Murali

, Barnhart

, Betthauser

, Gallagher

, Rowley

, Dowling

, Asthana

, Bendlin

, Blennow

, Zetterberg

, Carlsson

, Christian

, Johnson

(2016) Cerebrospinal fluid ratios with Aβ42 predict preclinical brain β-amyloid accumulation. Alzheimers Dement (Amst) 2, 27–38.

33.

Chen

, Pounds

(1998) A nonlinear isobologram model with Box-Cox transformation to both sides for chemical mixtures. Environ Health Perspect 106(Suppl 6), 1367–1371.

34.

Preacher

, Hayes

(2004) SPSS and SAS procedures for estimating indirect effects in simple mediation models. Behav Res Methods Instrum Comput 36, 717–731.

35.

Sobel

(1982) Asymptotic confidence intervals for indirect effects in structural equation models. Sociol Methodol 13, 290.

36.

Stephen

, Ngandu

, Liu

, Peltonen

, Antikainen

, Kemppainen

, Laatikainen

, Lötjönen

, Rinne

, Strandberg

, Tuomilehto

, Vanninen

, Soininen

, Kivipelto

, Solomon

(2021) Change in CAIDE dementia risk score and neuroimaging biomarkers during a 2-year multidomain lifestyle randomized controlled trial: Results of a post-hoc subgroup analysis. J Gerontol A Biol Sci Med Sci 76, 1407–1414.

37.

Hooshmand

, Polvikoski

, Kivipelto

, Tanskanen

, Myllykangas

, Mäkelä

, Oinas

, Paetau

, Solomon

(2018) CAIDE Dementia Risk Score, Alzheimer and cerebrovascular pathology: A population-based autopsy study. J Intern Med 283, 597–603.

38.

Stephen

, Liu

, Ngandu

, Rinne

, Kemppainen

, Parkkola

, Laatikainen

, Paajanen

, Hänninen

, Strandberg

, Antikainen

, Tuomilehto

, Keinänen Kiukaanniemi

, Vanninen

, Helisalmi

, Levälahti

, Kivipelto

, Soininen

, Solomon

(2017) Associations of CAIDE Dementia Risk Score with MRI, PIB-PET measures, and cognition. J Alzheimers Dis 59, 695–705.

39.

Enache

, Solomon

, Cavallin

, Kåreholt

, Kramberger

, Aarsland

, Kivipelto

, Eriksdotter

, Winblad

, Jelic

(2016) CAIDE Dementia Risk Score and biomarkers of neurodegeneration in memory clinic patients without dementia. Neurobiol Aging 42, 124–131.

40.

Fagan

(2015) What does it mean to be ‘amyloid-positive’? . Brain 138, 514–516.

41.

Liu

, Tan

, Zhang

, Yi

, Zhu

, Peng

(2024) The interaction between ageing and Alzheimer’s disease: Insights from the hallmarks of ageing. Transl Neurodegener 13, 7.

42.

Iadecola

, Gottesman

(2019) Neurovascular and cognitive dysfunction in hypertension. Circ Res 124, 1025–1044.

43.

Weller

, Boche

, Nicoll

JAR

(2009) Microvasculature changes and cerebral amyloid angiopathy in Alzheimer’ s disease and their potential impact on therapy. Acta Neuropathol 118, 87–102.

44.

Zlokovic

(2011) Neurovascular pathways to neurodegeneration in Alzheimer’s disease and other disorders. Nat Rev Neurosci 12, 723–738.

45.

Tesseur

, Van Dorpe

, Spittaels

, Van den Haute

, Moechars

, Van Leuven

(2000) Expression of human apolipoprotein E4 in neurons causes hyperphosphorylation of protein tau in the brains of transgenic mice. Am J Pathol 156, 951–964.

46.

Nelson

, Alafuzoff

, Bigio

, Bouras

, Braak

, Cairns

, Castellani

, Crain

, Davies

, Del Tredici

, Duyckaerts

, Frosch

, Haroutunian

, Hof

, Hulette

, Hyman

, Iwatsubo

, Jellinger

, Jicha

, Kövari

, Kukull

, Leverenz

, Love

, Mackenzie

, Mann

, Masliah

, McKee

, Montine

, Morris

, Schneider

, Sonnen

, Thal

, Trojanowski

, Troncoso

, Wisniewski

, Woltjer

, Beach

(2012) Correlation of Alzheimer disease neuropathologic changes with cognitive status: A review of the literature. J Neuropathol Exp Neurol 71, 362–381.

47.

Coradduzza

, Sedda

, Cruciani

, De Miglio

, Ventura

, Nivoli

, Maioli

(2023) Age-related cognitive decline, focus on microbiome: A systematic review and meta-analysis. Int J Mol Sci 24, 13680.

48.

Numakawa

, Odaka

(2022) The role of neurotrophin signaling in age-related cognitive decline and cognitive diseases. Int J Mol Sci 23, 7726.

49.

Lövdén

, Fratiglioni

, Glymour

, Lindenberger

, Tucker-Drob

(2020) Education and cognitive functioning across the life span. Psychol Sci Public Interest 21, 6–41.

50.

Stern

(2012) Cognitive reserve in ageing and Alzheimer’s disease. Lancet Neurol 11, 1006–1012.

51.

Ungvari

, Toth

, Tarantini

, Prodan

, Sorond

, Merkely

, Csiszar

(2021) Hypertension-induced cognitive impairment: From pathophysiology to public health. Nat Rev Nephrol 17, 639–654.

52.

Hötting

, Röder

(2013) Beneficial effects of physical exercise on neuroplasticity and cognition. Neurosci Biobehav Rev 37, 2243–2257.

53.

Hawkins

, Gathright

, Gunstad

, Dolansky

, Redle

, Josephson

, Moore

, Hughes

(2014) The MoCA and MMSE as screeners for cognitive impairment in a heart failure population: A study with comprehensive neuropsychological testing. Heart Lung 43, 462–468.